Atmos. Chem. Phys., 6, 93–108, 2006
www.atmos-chem-phys.org/acp/6/93/
SRef-ID: 1680-7324/acp/2006-6-93
European Geosciences Union
Atmospheric
Chemistry
and Physics
Intercomparison exercise between different radiative transfer
models used for the interpretation of ground-based zenith-sky and
multi-axis DOAS observations
F. Hendrick1 , M. Van Roozendael1 , A. Kylling2,* , A. Petritoli3 , A. Rozanov4 , S. Sanghavi5 , R. Schofield6,** , C. von
Friedeburg5 , T. Wagner5 , F. Wittrock4 , D. Fonteyn1 , and M. De Mazière1
1 Institut
d’Aéronomie Spatiale de Belgique, Brussels, Belgium
Institute for Air Research, Kjeller, Norway
3 Institute of Atmospheric Science and Climate, Bologna, Italy
4 Institute of Environmental Physics, University of Bremen, Bremen, Germany
5 Institute of Environmental Physics, University of Heidelberg, Heidelberg, Germany
6 National Institute of Water and Atmospheric Research, Omakau, Central Otago, New Zealand
* now at: St. Olavs University Hospital, Trondheim, Norway
** now at: NOAA Aeronomy Laboratory, Boulder, Colorado, USA
2 Norwegian
Received: 30 June 2005 – Published in Atmos. Chem. Phys. Discuss.: 2 September 2005
Revised: 29 November 2005 – Accepted: 13 December 2005 – Published: 20 January 2006
Abstract. We present the results of an intercomparison exercise between six different radiative transfer (RT) models carried out in the framework of QUILT, an EU funded project
based on the exploitation of the Network for the Detection of
Stratospheric Change (NDSC). RT modelling is an important
step in the interpretation of Differential Optical Absorption
Spectroscopy (DOAS) observations. It allows the conversion
of slant column densities (SCDs) into vertical column densities (VCDs) using calculated air mass factors (AMFs). The
originality of our study resides in comparing SCD simulations in multi-axis (MAX) geometry (trace gases: NO2 and
HCHO) and in taking into account photochemical enhancement for calculating SCDs of rapidly photolysing species
(BrO, NO2 , and OClO) in zenith-sky geometry. Concerning
the zenith-sky simulations, the different models agree generally well, especially below 90◦ SZA. At higher SZA, larger
discrepancies are obtained with relative differences ranging
between 2% and 14% in some cases. In MAX geometry,
good agreement is found between the models with the calculated NO2 and HCHO SCDs differing by no more than 5%
in the elevation and solar zenith angle (SZA) ranges investigated (5◦ –20◦ and 35◦ –85◦ , respectively). The impacts of
aerosol scattering, ground albedo, and relative azimuth on
MAX simulations have also been tested. Significant discrepancies appear for the aerosol effect, suggesting differences
between models in the treatment of aerosol scattering. A betCorrespondence to: F. Hendrick
([email protected])
ter agreement is found in case of the ground albedo and relative azimuth effects. The complete set of initialization data
and results have been made publicly available through the
QUILT project web site (http://nadir.nilu.no/quilt/), enabling
the testing of other RT codes designed for the calculation of
SCDs/AMFs.
1
Introduction
Since the middle of the 1970s, stratospheric ozone and several trace gases directly or indirectly involved in the ozone
depletion like NO2 , BrO, and OClO have been monitored
from the ground using the Differential Optical Absorption
Spectroscopy (DOAS)-technique (Platt, 1994). A significant
part of this monitoring effort has been carried out through
the framework of the Network for the Detection of Stratospheric Change (NDSC). The NDSC consists of about 75
globally distributed stations combining various observation
techniques. The network operation started in January 1991
and has been providing a consistent, standardised set of longterm measurements of stratospheric and, more recently, tropospheric trace gases, particles, and physical parameters for
detecting atmospheric change, validating space-borne sensors, and testing and improving multidimensional models of
both the stratosphere and troposphere (further information at
http://www.ndsc.ncep.noaa.gov/). Concerning the scattered
sunlight DOAS instruments, they operated until recently
© 2006 Author(s). This work is licensed under a Creative Commons License.
94
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
almost only in zenith-sky geometry, thus probing mainly the
altitude region corresponding to the stratosphere, especially
at large solar zenith angles (SZAs). Over the last decade,
new instruments pointing not only at zenith but also towards
the horizon (off-axis geometry) have been developed (e.g.,
Heckel et al., 2005; Hönninger et al., 2004; Wagner et al.,
2004; Wittrock et al., 2004). Pointing at an elevation angle
close to the horizon provides enhanced sensitivity to the troposphere compared to zenith-sky observations. The MultiAXis (MAX-) DOAS technique combines different elevation
angles. Owing to the variation of the light paths in the troposphere with the elevation angle, the different viewing directions have maximum sensitivity at different altitudes, thus
providing some information on the vertical distribution of the
absorber.
An important step in the interpretation of DOAS observations is the conversion of the slant column densities (SCDs)
– which are the direct product of the DOAS analysis – into
vertical column densities (VCDs) using calculated air mass
factors (AMFs). The AMF represents the ratio of the effective optical path through the atmosphere to the vertical optical path and is given by the ratio SCD/VCD (Noxon et al.,
1979; Solomon et al., 1987). In practice, AMFs are calculated using a radiative transfer (RT) model. In case of short
lived species such as BrO, OClO, and NO2 , the calculation
of SCDs and therefore AMFs is complicated by the variation of the concentration of these species along a given light
path due to the fast diurnal variation of these radicals coupled to the local SZA variation along the light path (Fish et
al., 1995). This so-called photochemical enhancement effect
is taken into account by initializing RT models with a table
containing the concentrations of the absorber for different altitudes and SZAs. This table is then interpolated to determine
the trace gas concentrations at the altitudes and local SZAs of
the different scatter points considered along a given sun ray
path. Such a concentration table is usually generated by running a stacked box photochemical model initialized with the
output of a 3-D chemical transport model (CTM) corresponding to the day and location of interest. SCDs calculated by
coupling a 3-D CTM to a photochemical box-/RT model interface can also be directly compared to observed SCDs. For
example, Sinnhuber et al. (2002) compared ground-based
zenith-sky UV-visible observations of BrO obtained at 11
sites with simulations from the 3-D CTM SLIMCAT (Chipperfield, 1999). The RT model used was based on the single scattering ray tracing scheme of Solomon et al. (1987).
Recently, several groups have initiated AMF calculations in
order to interpret MAX-DOAS measurements. Hönninger et
al. (2004) have studied the behaviour of AMFs as a function of the solar zenith, elevation, and relative azimuth angles, ground albedo, and aerosol loading for several idealized trace gas profiles. For this purpose, they have used the
RT model TRACY based on a Monte Carlo approach (von
Friedeburg, 2003). This model has been also used by Wagner
et al. (2004) to perform RT modelling for MAX-DOAS O4
Atmos. Chem. Phys., 6, 93–108, 2006
observations at different aerosol conditions. Sensitivity tests
on O4 AMFs were carried out by Wittrock et al. (2004) with
SCIATRAN, a RT code based on a combined differentialintegral approach involving the Picard iterative approximation (CDIPI) (Rozanov et al., 2000, 2001, 2005) and, as a
first application, NO2 profile information were derived from
MAX-DOAS observations. It should be noted that an accurate RT model as forward model is essential in the retrieval
of tropospheric trace gas profiles using MAX-DOAS measurements (e.g., Bruns et al., 2004). In Heckel et al. (2005),
the SCIATRAN model also generated appropriate AMFs for
the conversion of MAX-DOAS measurements of HCHO into
VCDs. These examples show that RT modelling plays a central role in the interpretation of ground-based DOAS observations and that different RT computation schemes are available for this purpose.
The QUILT (Quantification and Interpretation of LongTerm UV-Vis Observations of the Stratosphere) project is an
EU funded project based on the exploitation of the NDSC
and aimed at quantifying ozone loss and investigating its relation to active halogen and nitrogen species using the existing ground-based, satellite and balloon borne UV-visible
data as well as 3-D atmospheric modelling tools. One of
the tasks of this project has been to test the consistency between the different RT models existing within the consortium
and used to interpret the ground-based and satellite DOAS
data. This objective has been achieved through several SCD
simulations comparison tests performed using identical settings for all the models. Both ground-based zenith-sky (trace
gases: BrO, NO2 , and OClO) and MAX geometries (trace
gases: NO2 and HCHO) have been considered for these tests,
with photochemical enhancement being taken into account
only in zenith-sky geometry.
Here we report on the results of this intercomparison exercise. It should be noted that our study does not address
the issue of the absolute accuracy of the SCD calculations,
the consistency between simulated and measured SCDs having already been tested in several papers (e.g., Sinnhuber et
al., 2002; Tørnkvist et al., 2002). It can be considered to a
certain extent as the continuation of Sarkissian et al. (1995)
who calculated O3 and NO2 AMFs with different RT models in zenith-sky geometry but without taking into account
the photochemical enhancement effect (use of a single profile for the initialization of the models). Our paper is divided
into five parts. In Sect. 2, we describe the different RT models involved in the intercomparison exercise. The comparison tests in zenith-sky and MAX geometries are described in
Sects. 3 and 4, respectively, and their corresponding results
are discussed therein. Sects. 5 and 6 are dedicated to the
impact of aerosols and ground albedo on MAX simulations,
respectively.
www.atmos-chem-phys.org/acp/6/93/
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
95
Table 1. Short description of the RT models involved in the intercomparison exercise. Note that both IASB and NILU models are based on
the UVspec/DISORT package. However, they are not strictly identical since they have been adapted separately by each group in order to
allow SCD/AMF calculations taking into account photochemical enhancement.
Group
Model main features
UBRE
SCIATRAN model:
–Combining Differential-Integral approach using the
Picard-Iterative approximation (CDIPI)
–Treatment of MS in a full spherical geometry including refraction
–Chemistry included (2-D array of profile variation with SZA)
–Treatments for aerosol and cloud scattering, and ground albedo
–Possibility to include Raman scattering (but not used in the present study)
SS+MS
UHEI
TRACY model:
–Backward Monte Carlo approach
–Treatment of MS in a full spherical geometry including refraction
–Chemistry included (2-D array of profile variation with SZA)
–Treatments for aerosol and cloud scattering, and ground albedo
–Possibility to include Raman scattering (but not used in the present study)
SS+MS
(only MS
in this study)
NILU and
IASB
UVspec/DISORT package:
–Discrete ordinate method
–Chemistry included (2-D array of profile variation with SZA)
–Treatment of MS and refraction in a pseudo-spherical geometry
(direct beam only)
–Treatments for aerosol and cloud scattering, and ground albedo
ISAC-CNR
AMEFCO model:
–Single scattering model in a 2-D atmosphere
(profile variation with SZA)
–Treatment in full spherical geometry including refraction
and aerosol scattering
SS
Petritoli et al. (2002a, b)
NIWA
–Single scattering model in a 2-D atmosphere
(profile variation with SZA)
–Treatment in full spherical geometry including refraction
and aerosol scattering
SS
Schofield (2003);
Schofield et al. (2004)
2 Description of the RT models
The groups contributing RT calculations to the intercomparison exercise were the remote sensing groups at the Universities of Bremen (UBRE) and Heidelberg (UHEI), the Norwegian Institute for Air Research (NILU), the Institute of Atmospheric Science and Climate (ISAC-CNR), the National
Institute of Water and Atmospheric Research (NIWA), and
the Institut d’Aéronomie Spatiale de Belgique (IASB). A
summary of the characteristics of the models used by these
groups is given in Table 1. From now on, each model will
be referred to by the acronym of its corresponding group (in
brackets here above). All models include the possibility of
taking into account photochemical enhancement. The solution approaches to the RT equation (RTE) used in the different RT codes are as follows:
• The Combined Differential-Integral (CDI) approach
(UBRE model): is based on the solution of the integral
www.atmos-chem-phys.org/acp/6/93/
Scattering mode(s)
SS+MS
Reference(s)
Rozanov et al. (2000, 2001, and 2005);
Wittrock et al. (2004)
von Friedeburg (2003);
Hönninger et al. (2004)
Mayer and Kylling (2005)
RT equation using the characteristics method (Currant
and Hilbert, 1962). The source function is integrated
along the line of sight intersecting a spherical atmosphere. The single scattering part of the source function
is calculated truly spherical and the multiple scattering
part is initialized by the output of the pseudo-spherical
model (Rozanov et al., 2000 and 2001).
• The Discrete Ordinate method (NILU and IASB models): the azimuth dependence of the radiation field is
expressed as a Fourier cosine series in azimuth. The
solution of the Fourier components is obtained using a
numerical quadrature scheme, allowing to replace the
integrals by sums and thus reducing the RTE to the solution of a set of coupled linear first-order differential
equations (Lenoble, 1985; Stamnes et al., 1988; Spurr,
2001).
• The backward Monte Carlo method (UHEI model): a
photon emerges from a detector in an arbitrary line of
Atmos. Chem. Phys., 6, 93–108, 2006
96
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
Fig. 1. Contour plot of the BrO profiles table used to initialize the
models in the comparison test in zenith-sky geometry. The variation
of the BrO profiles as a function of the SZA has been calculated for
Harestua (60◦ N, Norway) at sunset under chlorine activated conditions.
Fig. 3. Contour plot of the NO2 profiles table used to initialize the
models in the comparison test in zenith-sky geometry. The variation
of the NO2 profiles as a function of the SZA has been calculated for
Harestua (60◦ N, Norway) at sunset in summer.
is calculated and weighted with the value of the scattering phase function and with the attenuation of the complete path. A large number of such random photon paths
will reproduce the light contributing to the simulated
measurement (von Friedeburg, 2003; Lenoble, 1985).
• Single scattering ray tracing method (ISAC-CNR and
NIWA models): in zenith-sky geometry, it consists of
computing the attenuation from the sun through discrete
spherical atmospheric shells to “points” where the light
beams are scattered into the detector (Solomon et al.,
1987).
3
3.1
Fig. 2. Contour plot of the OClO profiles table used to initialize the
models in the comparison test in zenith-sky geometry. The variation of the OClO profiles as a function of the SZA has been calculated for Harestua (60◦ N, Norway) at sunset under chlorine activated conditions.
sight direction and is followed in the backward direction
along the path towards the sun. The various events
which may happen to the photon at various altitudes in
the atmosphere are defined by suitable probability distributions. Random numbers decide on the occurrence
of events. At the location of the last scattering event
prior to leaving the atmosphere, the impinging radiance
Atmos. Chem. Phys., 6, 93–108, 2006
Zenith-sky simulations of BrO, NO2 , and OClO SCDs
including photochemical enhancement
Comparison test description
In this test, BrO, OClO, and NO2 SCDs have been calculated in zenith-sky geometry in single scattering (SS) and,
when possible, in multiple scattering (MS) modes, taking
into account photochemical enhancement effect. The settings imposed for all models are summarized in Table 2.
Contour plots of the diurnal variation tables (concentration
of BrO, OClO, and NO2 as a function of altitude and SZA)
are shown in Figs. 1, 2, and 3, respectively. They are the
output of the stacked box photochemical model PSCBOX
(Errera and Fonteyn, 2001; Hendrick et al., 2000 and 2004)
initialized with the 3-D-CTM SLIMCAT chemical and meteorological fields. Two scenarios have been considered:
Harestua (60◦ N, Norway) at sunset in summer for NO2 SCD
www.atmos-chem-phys.org/acp/6/93/
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
97
Table 2. Model settings for the comparison test in zenith-sky geometry. All the initialization data are available on the QUILT web site
(http://nadir.nilu.no/quilt/).
Altitude grid
Wavelength
Diurnal variation table (see Figs. 1, 2, and 3)
p, T profiles
O3 profile
NO2 profile (for BrO and OClO SCD calculations)
Mie scattering
Refraction
Cross sections sets
Ground albedo
Scattering mode
calculations and Harestua at sunset under chlorine activated
conditions for BrO and OClO. All calculations included absorption by O3 and, in case of BrO and OClO, also absorption by NO2 .
3.2
Results
The results of the zenith-sky simulations of BrO, NO2 , and
OClO SCDs are presented in Fig. 4. We first concentrate
on the general behaviour of the simulated SCDs. Figure 4
shows that the impact of MS is more important for BrO
than for NO2 , and for OClO, it is mostly significant at large
SZA (>92◦ ). The different behaviours of BrO and NO2 regarding MS effects can be mainly attributed to the different
wavelengths used for both calculations (352 nm for BrO and
422 nm for NO2 ). In case of OClO, the large impact of MS
at SZAs larger than ∼92◦ is related to the photochemistry of
this species combined with geometrical considerations. Figure 2 shows that OClO is situated in the ∼17–22 km altitude
range and only appears at SZAs>92◦ . At these SZAs, the
absorption due to OClO along a given line of sight occurs
where the local SZA is largest, i.e. near the vertical above
the observing point. Thus, a significant part of the OClO
layer is likely to be in the Earth’s shadow region where multiply scattered light has a stronger relative impact. This explains why the OClO SCDs are so sensitive to MS in these
conditions. The differences between the photochemical behaviours of BrO, NO2 , and OClO are also manifest from
www.atmos-chem-phys.org/acp/6/93/
0–120 km/1 km thick
352 nm (BrO)
368 nm (OClO)
422 nm (NO2 )
From PSCBOX model
From SLIMCAT 3-D-CTM; interpolation
using AFGL 1976 outside the SLIMCAT
altitude range
From PSCBOX model; interpolation
using AFGL 1976 outside the SLIMCAT
altitude range
From PSCBOX model; interpolation
using AFGL 1976 outside the SLIMCAT
altitude range
Not included
Not included
BrO: Wahner et al. (1988)
OClO: Wahner et al. (1987)
NO2 : Burrows et al. (1998)
O3 : Burrows et al. (1999)
0.20
SS+if possible MS
Fig. 4. BrO SCDs show a sharp decrease above ∼92◦ SZA
(see Fig. 1), consistent with the rapid conversion of BrO into
its night-time reservoirs (mostly BrCl in the present case) in
the absence of sunlight. In contrast, the concentrations of
OClO and NO2 increase with SZA (see Figs. 2 and 3, respectively), which explains the persistence of relatively large
SCD values at corresponding SZAs, especially in MS mode.
However, OClO and NO2 SCDs also decrease at very high
SZAs. This decrease happens when the mean scattering altitude moves above the trace gas layer. The SZA where the
maximum SCD is reached is thus determined by a combination of photochemistry and altitude range of the trace gas
layer.
Concerning the comparison of simulated SCDs, the relative differences between the results of the different models and those of IASB arbitrarily taken as reference, appear in the lower plots of Fig. 4 and a summary of the
maximum differences found is presented in Table 3. In SS
mode, good agreement is obtained below 90◦ SZA between
the NIWA, IASB, NILU, and UBRE models: for the three
species, the relative difference between the NIWA, NILU,
and UBRE models and the IASB one (reference model) is
smaller than 1%. However, the SCDs simulated by the ISACCNR model are systematically larger and the relative differences rise up to 4% for BrO, 5% for NO2 and 7.5% for
OClO. Above 90◦ SZA, the agreement between the NIWA,
IASB, UBRE, and NILU models is still very good with relative differences smaller than 1.8% for BrO, 1.3% for NO2 ,
Atmos. Chem. Phys., 6, 93–108, 2006
98
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
NO2
BrO
Harestua − 28/01/00 − PM
OClO
Harestua − 19/06/99 − PM
Harestua − 28/01/00 − PM
12
3
2.5
2
1.5
1
8
6
3.5
4
2
0.5
0
74
10
4
IASB SS
IASB MS
NILU SS
NILU MS
NIWA SS
ISAC SS
UBRE SS
UBRE MS
UHEI MS
OClO SCD [x1014 molec/cm2]
3.5
NO2 SCD [x1016 molec/cm2]
BrO SCD [x1014 molec/cm2]
4
78
82
86
SZA [°]
90
0
74
94
2.5
2
1.5
1
0.5
78
82
86
SZA [°]
90
94
15
15
3
0
74
5
5
0
0
0
−5
−5
−5
Relative difference [%]
10
78
82
86
SZA [°]
90
94
−10
74
82
86
SZA [°]
90
94
78
82
86
SZA [°]
90
94
15
NILU SS−IASB SS
NIWA SS−IASB SS
ISAC SS−IASB SS
UBRE SS−IASB SS
NILU MS−IASB MS
UBRE MS−IASB MS
UHEI MS−IASB MS
10
−10
74
78
78
82
86
SZA [°]
90
10
5
94
−10
74
Fig. 4. Sunset BrO, NO2 , and OClO SCDs calculated in the comparison test in zenith-sky geometry. The upper plots correspond to the
SCDs and the lower plots to the relative differences between the results from the different models and those from the IASB model arbitrarily
chosen as reference. Solid and dashed lines correspond to calculations in SS and MS modes, respectively. Note that the UHEI OClO data are
missing (see Sect. 3.2) and in the upper plots, the blue, red, and cyan solid lines corresponding to, respectively, the IASB, NILU, and NIWA
SCD calculations in SS mode are almost superimposed.
Table 3. Maximum relative differences obtained between the different RT models and the IASB one – arbitrarily taken as reference
– for the comparison test in zenith-sky geometry. Note that the SS
and MS models have been compared to the IASB model in SS and
MS modes, respectively.
NILU SS
ISAC SS
NIWA SS
UBRE SS
NILU MS
UHEI MS
UBRE MS
BrO [%]
NO2 [%]
OClO [%]
+1.0
+4.1
+1.5
+1.8
+5.8
−5.0
−1.0
+0.5
+7.1
+0.7
+1.3
+2.7
+5.5
−0.6
+2.1
+14.3
+1.6
−1.4
+6.6
–
−4.2
Atmos. Chem. Phys., 6, 93–108, 2006
and 2.1% for OClO. A maximum spread between models of
about 2% is a better agreement than obtained by Sarkissian
et al. (1995) without taking into account photochemical enhancement (from Sarkissian et al. (1995), the maximum
spread values rise up to 13% for O3 and 8% for NO2 at
94◦ SZA). Concerning the ISAC-CNR simulations above 90◦
SZA, the discrepancies with the IASB model are larger than
for the NIWA and NILU models: the relative differences are
smaller than 4% for BrO and rise up to 7.1% for NO2 , and
14.3% for OClO. Although these maximum relative differences are consistent with those of Sarkissian et al. (1995),
they suggest that the ISAC-CNR model differs from the other
models in the way the optical thickness of each atmospheric
shell is calculated. Sarkissian et al. (1995) investigated the
influence of the different methods used to interpolate the concentration and to calculate the geometrical path in individual shells. They found that the impact of these parameters
www.atmos-chem-phys.org/acp/6/93/
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
increases with increasing SZA and can reach 4% above 90◦
SZA. Since photochemical enhancement is included in the
present test, we cannot neglect the impact of differences in
interpolating the tables of photochemical species concentrations. This effect is consistent with the increase in the relative
differences between the models with increasing SZA since
the step between the SZA values corresponding to the concentration tables is increasing at large SZAs, due to the fact
that the stacked box photochemical model PSCBOX provides output with a constant time step. Concerning the calculations in MS mode, the agreement below 90◦ SZA between the NILU, IASB, UBRE, and UHEI models is good
since the relative differences with the IASB model results
are smaller than 4.4%. Above 90◦ SZA, slightly larger discrepancies are generally found, as in SS mode, and the maximum relative differences with the IASB model results rise
up to 6.6% for the NILU model (obtained for OClO SCD
calculations), 5.5% for the UHEI model (obtained for NO2
SCD calculations), and 4.2% for the UBRE model (obtained
for OClO SCD calculations). It should be noted that UHEI
OClO data are absent from the comparison. This is caused
by the fact that too many photons (around 105 ) would have
been required to calculate each SCD with sufficient accuracy
and precision. In contrast to the NO2 or BrO profile, stratospheric OClO is usually confined to a layer of only a few
kilometres thickness, and the concentration gradient is very
high at its lower and upper ends (see Fig. 2). The Monte
Carlo nature of the UHEI model causes statistical fluctuations (from photon path to photon path) of the altitude of the
first scattering event the photon encounters after atmosphere
entry. If this event occurs above the OClO layer, then the
path through this layer is short. If the event occurs within the
OClO layer, the light path through this layer is long. If the
event occurs below the OClO layer, then the path through the
layer has also been long, though shorter than in the second
case. Accordingly, small variations in the altitude of the first
scattering event, arising from statistical fluctuations, lead to
large variations in path length and hence to large fluctuations
in the simulated SCD. This noise can only be reduced with
a very large number of modelled photons, which causes the
UHEI model to be inefficient for this specific kind of scenario. With the expected increase in computational power,
this disadvantage is likely to become less significant in the
near future.
4 Multi-axis simulations of NO2 and HCHO SCDs
4.1
Comparison test description
In contrast to the zenith-sky simulations, photochemical enhancement has not been taken into account in the MAX simulations. The reason is that this effect becomes significant
only during twilight, i.e. above 85◦ SZA, and we have limited
our comparison to the 38◦ –85◦ SZA range because the senwww.atmos-chem-phys.org/acp/6/93/
99
sitivity of the MAX-DOAS observations to the troposphere
is largest during daytime, i.e. at small SZA. The NO2 and
HCHO profiles used for the initialization of the models are
shown in Fig. 5. All calculations included absorption by O3
and in case of HCHO, also absorption by NO2 . SCD simulations have been performed for 5◦ , 10◦ , and 20◦ elevation
above the horizon. For each elevation angle, four values of
the azimuth angle of line of sight (AzLOS) have been considered: 30◦ , 60◦ , 90◦ , and 120◦ . The AzLOS is the relative
azimuth angle between the pointing direction of the instrument and the sun azimuth. The other settings are shown in
Table 4. Concerning the HCHO SCD calculations, the layer
thickness in the altitude range where the concentration of this
species changes rapidly with altitude (0–3 km) has been fixed
to 0.2 km instead of the 1 km thickness used in case of NO2 ,
reducing interpolation effects. Only models enabling calculations in MS mode have been involved in this test since
MS is the only acceptable scattering mode in MAX geometry (Wittrock et al., 2004). These models are: IASB, NILU,
UBRE and UHEI.
4.2
Results
Figure 6 shows the results of the NO2 and HCHO SCD simulations for 5◦ , 10◦ , and 20◦ elevation and 90◦ AzLOS. A
summary of the maximum differences between the different
models and the IASB one (again taken arbitrarily as reference model) is also presented in Table 5. The agreement
between IASB, NILU, UBRE and UHEI is good: when comparing the results for both species over the whole SZA range
and for all the elevation angles, the relative differences between the NILU, UBRE, and UHEI models and the IASB
one are smaller than 5.1%. Concerning the UBRE model, it
underestimates the IASB model results by at most 4.5% for
NO2 and 5.1% for HCHO. Furthermore, the relative differences are also almost constant over the whole SZA range.
This is in contrast to the relative differences between the
UHEI and IASB models which show a SZA dependence,
especially at low elevation angles: e.g., for NO2 at 5◦ elevation, the UHEI model results overestimate the IASB ones
by 3% at 40◦ SZA and underestimate them by 3.6% at 85◦
SZA. However, with maximum relative differences of 3.8%
for NO2 and 4.5% for HCHO (see Table 5), the agreement
between both models is good. As expected since they are
both based on the UVspec/DISORT package, the NILU and
the IASB models agree very well: for both NO2 and HCHO,
the relative difference between the results of both models is
smaller than 1.3% over the entire SZA range and for all elevation angles.
The impact of the relative azimuth has been tested by calculating NO2 and HCHO SCDs for different AzLOS values
(30◦ , 60◦ , 90◦ , and 120◦ ). The AzLOS effect on NO2 SCDs
is illustrated in Fig. 7 where the relative differences between
the NO2 SCDs at 30◦ , 60◦ , and 120◦ AzLOS and the SCD
at 90◦ AzLOS (reference) are plotted as a function of SZA
Atmos. Chem. Phys., 6, 93–108, 2006
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
Altitude [km]
100
60
60
50
50
40
40
30
30
20
20
10
10
Altitude [km]
0
0
0
0
1
2
3
4
5
NO2 conc. [x109 molec/cm3]
60
60
50
50
40
40
30
30
20
20
10
10
0
0
1
2
3
NO2 VMR [ppbv]
4
0
0
20
40
60
HCHO conc. [x109 molec/cm3]
0.5
1
1.5
2
HCHO VMR [ppbv]
2.5
Fig. 5. NO2 (left) and HCHO (right) profiles in concentration (upper plots) and VMR (lower plots) used to initialize the RT models in the
comparison tests in MAX geometry.
Table 4. Model settings for the comparison test in MAX geometry. It should be noted that the same settings are used for the tests on aerosols
effect (except Mie scattering is then included; see Sect. 5) and ground albedo effect (with the ground albedo value then fixed to 0 and 0.9
instead of 0.2; see Sect. 6). All the initialization data are available on the QUILT web site (http://nadir.nilu.no/quilt/).
Altitude grid
Wavelength
p, T profiles
O3 profile
NO2 profile (for HCHO SCD calculations)
HCHO and NO2 profiles (see Fig. 5)
Mie scattering
Refraction
Cross sections sets
Ground albedo
Elevation angle
Azimuth angle of line of sight
Scattering mode
Atmos. Chem. Phys., 6, 93–108, 2006
NO2 : 0–120 km/1 km thick
HCHO: 0–120 km/0.2 km thick between
0 and 3 km and 1 km thick above 3 km
356 nm (HCHO)
422 nm (NO2 )
From SLIMCAT 3-D-CTM; interpolation using
AFGL 1976 outside the SLIMCAT altitude range
From PSCBOX model; interpolation using
AFGL 1976 outside the SLIMCAT altitude range
From PSCBOX model; interpolation using
AFGL 1976 outside the SLIMCAT altitude range
HCHO: Barbe et al. (1979);
Ehhalt and Tönnißen (1979)
NO2 : from PSCBOX model
Not included
Not included
HCHO: Cantrell et al. (1990)
NO2 : Burrows et al. (1998)
O3 : Burrows et al. (1999)
0.20
5◦ , 10◦ , and 20◦
30◦ , 60◦ , 90◦ , and 120◦
MS
www.atmos-chem-phys.org/acp/6/93/
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
101
NO2
90° AzLOS − 10° elev
NO2 SCD [x1016 molec/cm2]
90° AzLOS − 5° elev
6
6
6
4
4
4
2
2
2
40
60
SZA [°]
80
10
Relative difference [%]
90° AzLOS − 20° elev
40
60
SZA [°]
80
10
40
5
5
0
0
0
−5
−5
−5
40
60
SZA [°]
80
−10
60
SZA [°]
80
10
5
−10
IASB MS
NILU MS
UBRE MS
UHEI MS
40
60
SZA [°]
80
−10
NILU MS−IASB MS
UBRE MS−IASB MS
UHEI MS−IASB MS
40
60
SZA [°]
80
HCHO SCD [x1016 molec/cm2]
HCHO
90° AzLOS − 20° elev
6
6
6
4
4
4
2
2
2
40
60
SZA [°]
80
10
Relative difference [%]
90° AzLOS − 10° elev
90° AzLOS − 5° elev
40
60
SZA [°]
80
10
40
5
5
0
0
0
−5
−5
−5
40
60
SZA [°]
80
−10
60
SZA [°]
80
10
5
−10
IASB MS
NILU MS
UBRE MS
UHEI MS
40
60
SZA [°]
80
−10
NILU MS−IASB MS
UBRE MS−IASB MS
UHEI MS−IASB MS
40
60
SZA [°]
80
Fig. 6. NO2 and HCHO SCDs calculated in MS mode in the comparison test in MAX geometry. For each species, the upper plots correspond
to the SCDs and the lower plots to the relative differences between the different models and the IASB one arbitrarily chosen as reference.
Results for 90◦ AzLOS and for 5◦ , 10◦ , and 20◦ elevation are plotted here.
www.atmos-chem-phys.org/acp/6/93/
Atmos. Chem. Phys., 6, 93–108, 2006
102
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
Table 5. Maximum relative differences obtained between the different RT models and the IASB one – arbitrarily taken as reference – for the
comparison test in MAX geometry.
UHEI MS
UBRE MS
30° AzLOS
5° elev
IASB MS
NILU MS
UBRE MS
UHEI MS
Relative difference [%]; Reference: 90° AzLOS
+0.2
+1.3
−3.2
−3.0
−3.7
−5.0
+0.3
+1.1
−3.8
−4.5
−3.3
−5.1
ative azimuth effect is expected in presence of aerosols, due
to the strong preference of forward scattering by particles,
whereas the Rayleigh phase function is symmetric in forward and backward directions. Concerning the consistency
between the IASB, NILU, UBRE, and UHEI models with
respect to the impact of the AzLOS, Fig. 7 shows that they
agree reasonably well.
20° elev
−5
40
60
80
−10
5
5
0
0
−5
−5
40
60
80
−10
5
5
0
0
−5
−5
60
SZA [°]
80
−10
60
80
5
20° elev
40
120° AzLOS
5° elev
40
40
60° AzLOS
5° elev
−10
−0.2
+1.0
−3.6
−3.0
−4.5
−5.0
0
−5
−10
20◦ elevation [%]
5
0
−10
10◦ elevation [%]
NO2
HCHO
NO2
HCHO
NO2
HCHO
NILU MS
5
5◦ elevation [%]
60
80
20° elev
40
60
SZA [°]
80
Fig. 7. Impact of the relative azimuth on the NO2 SCDs calculated
in MAX geometry. The relative differences between the NO2 SCDs
calculated at 30◦ (upper plots), 60◦ (middle plots), and 120◦ (lower
plots) AzLOS and those calculated at 90◦ AzLOS (reference case)
are plotted here. Left and right plots correspond to results for 5◦
and 20◦ elevation, respectively. Note that the IASB blue lines are
superimposed to the NILU red lines.
for 5◦ and 20◦ elevation. The largest AzLOS impact on NO2
SCDs (7%) occurs at 50◦ SZA for 30◦ AzLOS and 20◦ elevation, i.e. at small AzLOS and high elevation angle. In case
of HCHO (not shown here), the AzLOS effect is smaller by
1–2% compared to NO2 . It should be noted that a larger relAtmos. Chem. Phys., 6, 93–108, 2006
Impact of aerosol scattering on multi-axis simulations
of NO2 and HCHO SCDs
Aerosols have a strong impact on MAX-DOAS observations
(Wagner et al., 2004; Wittrock et al., 2004; and Heckel et al.,
2005). The most important effect of the aerosol extinction
is a reduction of the visibility of the atmosphere and thus a
limitation by scattering of the light path of the lowest viewing
directions, reducing the difference in tropospheric absorption
path between the viewing directions. It has been shown by
Wagner et al. (2004) that the impact of aerosols on O4 SCDs
can provide a new method to derive information on atmospheric aerosols.
In the present study, MAX NO2 and HCHO SCDs have
been simulated with and without aerosol scattering in order
to test the consistency between the different RT models regarding the impact of aerosols in this geometry. The aerosol
settings are shown in Fig. 8 (profiles of the extinction and absorption coefficients) and Table 6 (asymmetry factor). They
have been constructed from the aerosol model of Shettle
(1989) included in the IASB and NILU RT models using a
surface visibility of 100 km. The settings in the boundary
layer and troposphere correspond to mixture of water soluble and dust-like aerosols representative of a rural environment (low aerosol content). In the stratosphere, sulphuric
acid aerosol settings corresponding to summer background
conditions have been used. The other initialization parameters are the same as described in Sect. 4.1, except that HCHO
and NO2 SCDs have been calculated only for 90◦ AzLOS.
Figure 9 shows MAX NO2 and HCHO SCDs simulated
with and without aerosol scattering. Qualitatively, the impact of aerosols is the same for all models: a decrease of NO2
and HCHO SCDs occurs when aerosol scattering is included
www.atmos-chem-phys.org/acp/6/93/
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
Extinction coefficient
Absorption coefficient
35
35
356 nm
422 nm
Altitude [km]
30
25
20
20
15
15
10
10
5
5
0.01
0.02 0.03
[km−1]
0.04
356 nm
422 nm
30
25
0
0
103
0
−10
10
−8
−6
10
10
[km−1]
−4
10
−2
10
Fig. 8. Profiles of the aerosol extinction (left) and absorption (right) coefficients used for testing the impact of aerosols on MAX simulations
of HCHO (356 nm) and NO2 (422 nm) SCDs, respectively.
in the calculations. The presence of aerosols leads to a reduction of the direct light path, especially in the lowest layers, and therefore the absorption by NO2 and HCHO is reduced (Wagner et al., 2004). The impact of aerosol scattering
also decreases at higher elevation angles, due to the fact that,
as the elevation angle increases, the mean scattering height
increases and therefore the MAX SCD simulations become
less sensitive to the lowest layers where the aerosol effect is
largest.
When SCDs calculated with aerosols are compared, the
different models agree reasonably well: they differ by 9% at
most, which is of the same order of magnitude as the differences found for calculations without aerosols (5%, see
Sect. 4.2). However, regarding the quantitative impact of the
aerosol scattering, expressed in Fig. 9 by the relative differences between simulations with and without aerosol scattering, the models differ significantly. For example, in HCHO
SCD simulations at 5◦ elevation, an aerosol impact of about
10% is found with the UHEI and UBRE models, which is
twice as much as the impact obtained with the NILU and
IASB models. In case of NO2 , such a difference is found
between the UBRE and UHEI model results (see results for
5◦ elevation in Fig. 9). Figure 9 also shows that the impact
of aerosol scattering is larger on NO2 than on HCHO SCDs:
the relative differences between calculations with and without aerosol scattering range for all models, elevation angles,
and SZAs between −21% and +4% for NO2 and between
−13% and +1.5% for HCHO. These discrepancies obtained
between the different models could not be resolved satisfactorily so far. It may be related to the different approximations
used in the models for resolving the RT equation. For examwww.atmos-chem-phys.org/acp/6/93/
Table 6. Profiles of asymmetry factor used for testing the impact of aerosols on MAX simulations of HCHO (356 nm) and NO2
(422 nm) SCDs.
0–10 km
11–30 km
>30 km
Asymmetry factor
356 nm
Asymmetry factor
422 nm
0.661
0.698
0.701
0.652
0.693
0.698
ple, the UBRE and UHEI models are spherical whereas the
NILU and IASB models use the pseudo-spherical approximation. A Fourier expansion is also applied in the NILU
and IASB models, which could lead to a lose of angular information if too few Fourier terms are considered. In order to progress further in our understanding of the persistent
discrepancies regarding the aerosol impact, a thorough examination of individual aerosol routines and additional comparison tests appear to be needed, which is beyond the scope of
the present intercomparison exercise. This issue, and more
generally the impact of aerosol scattering on MAX DOAS
AMF simulations, is addressed more thoroughly as part of
a new exercise, currently led by the University of Heidelberg in the framework of the European Network of Excellence on Atmospheric Composition Change ACCENT. More
details on this new exercise as well as preliminary results
can be found on the following web page: http://satellite.iup.
uni-heidelberg.de/index.php/RTM Workshop/149/0/.
Atmos. Chem. Phys., 6, 93–108, 2006
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
Relative difference [%]
NO2 SCD [x1016 molec/cm2]
104
90° AzLOS − 10° elev
90° AzLOS − 5° elev
8
8
90° AzLOS − 20° elev
8
6
6
6
4
4
4
2
2
2
40
60
SZA [°]
80
40
60
SZA [°]
80
40
5
5
5
0
0
0
−5
−5
−5
−10
−10
−10
−15
−15
−15
−20
−20
40
60
SZA [°]
80
IASB no aero
IASB aero
NILU no aero
NILU aero
UBRE no aero
UBRE aero
UHEI no aero
UHEI aero
60
SZA [°]
80
80
IASB aero−no aero
NILU aero−no aero
UBRE aero−no aero
UHEI aero−no aero
−20
40
60
SZA [°]
40
60
SZA [°]
80
Relative difference [%]
HCHO SCD [x1016 molec/cm2]
HCHO
90° AzLOS − 10° elev
90° AzLOS − 5° elev
8
8
90° AzLOS − 20° elev
8
6
6
6
4
4
4
2
2
2
40
60
SZA [°]
80
40
60
SZA [°]
80
40
5
5
5
0
0
0
−5
−5
−5
−10
−10
−10
−15
−15
−15
−20
−20
−20
40
60
SZA [°]
80
40
60
SZA [°]
80
IASB no aero
IASB aero
NILU no aero
NILU aero
UBRE no aero
UBRE aero
UHEI no aero
UHEI aero
60
SZA [°]
80
IASB aero−no aero
NILU aero−no aero
UBRE aero−no aero
UHEI aero−no aero
40
60
SZA [°]
80
Fig. 9. Impact of aerosol scattering on simulated MAX NO2 and HCHO SCDs. For each species, the upper plots correspond to the SCDs
and the lower plots to the relative differences between SCDs calculated with and without aerosol scattering (reference: without aerosol
scattering). Results for 90◦ AzLOS and for 5◦ , 10◦ , and 20◦ elevation are plotted here.
Atmos. Chem. Phys., 6, 93–108, 2006
www.atmos-chem-phys.org/acp/6/93/
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
6 Impact of ground albedo on multi-axis simulations of
NO2 and HCHO SCDs
Ground albedo has a significant impact on the radiative transfer close to the ground (Hönninger et al., 2004; Wittrock et
al., 2004). It is particularly important to properly estimate
this parameter for observational sites displaying large albedo
changes depending on the season or viewing direction. The
main effect of an increase of the albedo is an increase of the
number of scattering events in the layers close to the surface, resulting in longer absorption paths at these altitudes
and therefore in higher absolute AMFs. However, the difference in AMF between off-axis and zenith-sky viewing directions is reduced, owing to the fact that the enhancement of
the optical path with increasing albedo is largest for zenithsky observations (Wittrock et al., 2004).
In order to test the consistency between the different RT
models regarding the impact of the ground albedo in MAX
geometry, NO2 and HCHO SCDs have been simulated with
ground albedo values fixed to 0 and 0.9. The other model
settings are the same as those described in Sect. 4.1. As for
the test on the impact of aerosol scattering, simulations have
been performed only for 90◦ AzLOS.
Figure 10 illustrates the ground albedo effect on simulated
MAX HCHO and NO2 SCDs. The UBRE, NILU, and IASB
models show excellent consistency with similar HCHO and
NO2 SCD increases being obtained with the three models.
An increase in surface albedo value from 0.0 to 0.9 leads to
an increase of the HCHO SCD of about 20% and 55% at
80◦ SZA for 5◦ and 20◦ elevation, respectively, whereas the
NO2 SCD increases by 5% and 12% under the same conditions. The fact that the HCHO SCDs is more sensitive to the
ground albedo is expected since for the present simulations,
HCHO, in contrast to NO2 , is mostly located in the lower
part of the troposphere (between 0 and 5 km). In case of the
UHEI model, the SCDs simulated for a surface albedo value
of 0.0 agree reasonably well with the results from the other
models. This agreement is similar to the one obtained for
medium (0.2) surface albedo conditions (see Sect. 4). For
very high surface albedo (0.9), the UHEI model gives generally larger SCD values, especially above 70◦ SZA. However,
in case of HCHO SCDs, the differences between the UHEI
model results and the analytical models results are similar
to the differences found between the SCDs calculated by the
three analytical models. Due to the larger SCD values obtained when surface albedo is fixed to 0.9, the UHEI model
gives a significantly larger relative increase in SCDs compared to the other models: when the surface albedo increases
from 0.0 to 0.9, the SCD increases at 80◦ SZA for 5◦ and 20◦
elevation reach 32% and 84% for HCHO, and 12% and 35%
for NO2 . This behaviour could be related to the fact that the
modelling of the absolute radiance still poses a challenge to
backward Monte Carlo approaches, which is also under investigation by other groups. At high albedo, light reflected
off the ground is unattenuated and increases the signal of abwww.atmos-chem-phys.org/acp/6/93/
105
sorbers near the ground, so the absolute radiance begins to
play a significant role in the SCD calculations. The present
intercomparison exercise was aimed at detecting those subtle
effects, and the results will help to optimize the modelling.
So will comparison with measurements, e.g., as performed
in Weidner et al. (2005).
7
Conclusions
In the present intercomparison exercise, we have tested the
consistency between six RT models used for interpreting
ground-based zenith-sky and MAX-DOAS observations in
the QUILT EU project. In the context of this project based on
the exploitation of the NDSC, the comparison and optimization of these RT models is of central importance in order to
provide accurate time-series of ground-based DOAS observations. This study represents a step forward with respect
to previously published work in that it compares RT models in MAX geometry and takes into account photochemical enhancement effect for calculating SCDs of rapidly photolysing species in zenith-sky geometry.
Comparisons of NO2 and HCHO SCDs in MAX geometry and multiple scattering mode without aerosol scattering show good agreement between all involved models: the
calculated NO2 and HCHO SCDs differ generally by no
more than 5% in the SZA and elevation angle ranges investigated (35◦ –85◦ and 5◦ –20◦ , respectively). The impacts of
the relative azimuth, aerosol scattering, and ground albedo
on NO2 and HCHO SCDs have been also quantitatively determined. The azimuth effect is found to be the largest at
small AzLOS and high elevation angles. The maximum
differences relative to 90◦ AzLOS in the 30◦ –120◦ AzLOS
range and for 20◦ elevation rising up to 7% and 5% for
NO2 and HCHO SCDs, respectively. The models have also
shown reasonably good consistency concerning this effect.
This is in contrast to the aerosol scattering effect for which
significant discrepancies still persist. Since all codes have
been initialized identically, this result suggests that significant differences exist between the models regarding the treatment of aerosol scattering, may be related to the different
approximations used for resolving the RT equation. This
issue, and more generally the impact of aerosol scattering
on the MAX DOAS AMF simulations, is addressed more
thoroughly as part of a new intercomparison exercise, currently led by the University of Heidelberg in the framework
of the European Network of Excellence ACCENT. More
details on this new exercise as well as preliminary results
can be found on the following web page: http://satellite.iup.
uni-heidelberg.de/index.php/RTM Workshop/149/0/. In the
test on the impact of the ground albedo, very good agreement
has been achieved between the IASB, NILU, and UBRE
models whereas the albedo effect is significantly larger using
the UHEI model, which is based on a Monte Carlo approach.
It should be noted that the conclusions drawn here on the
Atmos. Chem. Phys., 6, 93–108, 2006
106
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
90° AzLOS − 10° elev
NO2 SCD [x1016 molec/cm2]
90° AzLOS − 5° elev
90° AzLOS − 20° elev
8
8
8
6
6
6
4
4
4
2
2
2
40
60
SZA [°]
80
40
60
SZA [°]
80
40
100
100
80
80
80
60
60
60
40
40
40
20
20
20
Relative difference [%]
100
0
40
60
SZA [°]
0
80
40
60
SZA [°]
80
IASB alb 0.0
IASB alb 0.9
NILU alb 0.0
NILU alb 0.9
UBRE alb 0.0
UBRE alb 0.9
UHEI alb 0.0
UHEI alb 0.9
0
60
SZA [°]
80
IASB alb 0.9−alb 0.0
NILU alb 0.9−alb 0.0
UBRE alb 0.9−alb 0.0
UHEI alb 0.9−alb 0.0
40
60
SZA [°]
80
90° AzLOS − 5° elev
8
IASB alb 0.0
IASB alb 0.9
NILU alb 0.0
NILU alb 0.9
UBRE alb 0.0
UBRE alb 0.9
UHEI alb 0.0
UHEI alb 0.9
4
2
40
60
SZA [°]
90° AzLOS − 20° elev
8
8
6
6
4
4
2
2
80
100
Relative difference [%]
90° AzLOS − 10° elev
6
HCHO SCD [x10
16
2
molec/cm ]
HCHO
40
60
SZA [°]
80
100
80
60
60
60
40
40
40
20
20
20
40
60
SZA [°]
80
0
60
SZA [°]
80
40
60
SZA [°]
80
100
IASB alb 0.9−alb 0.0
NILU alb 0.9−alb 0.0
UBRE alb 0.9−alb 0.0
UHEI alb 0.9−alb 0.0
80
0
40
40
60
SZA [°]
80
80
0
Fig. 10. Impact of the ground albedo on simulated MAX NO2 and HCHO SCDs. For each species, the upper plots correspond to the SCDs
and the lower plots to the relative differences between SCDs calculated with ground albedo values of 0.9 and 0.0 (reference: albedo=0.0).
Results for 5◦ , 10◦ , and 20◦ elevation and 90◦ AzLOS are plotted here.
Atmos. Chem. Phys., 6, 93–108, 2006
www.atmos-chem-phys.org/acp/6/93/
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
levels of agreement between models in the different comparison tests depend on the assumptions made for the vertical
profiles. In case of NO2 and HCHO, very different vertical
profiles are indeed possible and could lead to different levels
of agreement.
The comparisons of zenith-sky BrO, NO2 , and OClO
SCDs calculated in both single and multiple scattering modes
show good overall agreement to a level that is consistent and
in some cases better than in the Sarkissian et al. (1995) work.
In single scattering mode, the relative difference between the
models is smaller than 1% below 90◦ SZA, except for the
ISAC-CNR model,for which relative differences rising up to
7.5% have been found in this SZA range. Above 90◦ SZA,
the agreement between the IASB, NILU, UBRE, and NIWA
models is still very good with a maximum spread of about
2%. Larger discrepancies have been obtained with the ISACCNR model, especially in the OClO SCD calculations (relative difference with the IASB model up to 14%). These discrepancies could be partly due to differences in calculating
the density in individual atmospheric shells. In multiple scattering mode, the differences between the NILU, UHEI, and
UBRE models and the IASB one are smaller than 4.4% below 90◦ SZA for the three species. Above 90◦ SZA, slightly
larger discrepancies have been obtained with a maximum relative difference with the IASB model of about 6.5%. As in
MAX geometry, the assumptions made on the vertical profiles used for the zenith-sky simulations can have an impact
on the agreement found between the models. This is particularly true for OClO, for which a small change in the vertical
profiles could change the behaviour of the different models.
The complete set of initialization data (profiles and cross
sections sets) and results of this intercomparison exercise
have been gathered in a “RT model validation package” enabling the testing of other RT codes aiming to the calculation of SCDs/AMFs. This validation package has been
made publicly available through the QUILT project web site
(http://nadir.nilu.no/quilt/).
Acknowledgements. This research was financially supported by
the European Commission (contract QUILT, EVK2-2000-00545)
and the Belgian Federal Science Policy Office (contracts ESAC II
EV/35/3A and MO/35/006 & 012). M. P. Chipperfield (University
of Leeds) is acknowledged for providing us with the SLIMCAT
data. We would like also to thank the coordinators of the QUILT
EC project, B. Arlander and G. O. Braathen (NILU).
Edited by: U. Pöschl
References
Barbe, A., Marché, P., Secroun, C., and Jouve, P.: Measurements of
tropospheric and stratospheric H2 CO by an infrared high resolution technique, Geophys. Res. Lett., 6(6), 463–465, 1979.
Bruns, M., Buehler, S. A., Burrows, J. P., Heue, K.-P., Platt, U.,
Pundt, I., Richter, A., Rozanov, A., Wagner, T., and Wang, P.:
www.atmos-chem-phys.org/acp/6/93/
107
Retrieval of profile information from airborne multiaxis UVvisible skylight absorption measurements, Appl. Opt., 43(22),
2004.
Burrows, J. P., Dehn, A., Deters, B., Himmelmann, S., Richter,
A., Voigt, S., and Orphal, J.: Atmospheric remote-sensing reference data from GOME: Part 1. Temperature dependent absorption cross sections of NO2 in the 231–794 nm range, J. Quant.
Spectrosc. Radiat. Transfer, 60, 1025–1031, 1998.
Burrows, J. P., Richter, A., Dehn, A., Deters, B., Himmelmann,
S., Voigt, S., and Orphal, J.: Atmospheric remote-sensing reference data from GOME: Part 2. Temperature dependent absorption cross sections of O3 in the 231–794 nm range, J. Quant.
Spectrosc. Radiat. Transfer, 61, 509-517, 1999.
Cantrell, C. A., Davidson, J. A., McDaniel, A. H., Shetter, R. E.,
and Calvert, J. G.: Temperature-dependent formaldehyde cross
sections in the near-ultraviolet region, J. Phys. Chem., 94, 3902–
3908, 1990.
Chipperfield, M. P.: Multiannual simulations with a threedimensional chemical transport model, J. Geophys. Res.,
104(D1), 1781–1805, 1999.
Currant, R. and Hilbert, D.: Methods of Mathematical Physics, vol.
2, Interscience, New York, 1962.
Errera, Q. and Fonteyn, D.: Four-dimensional variational chemical assimilation of CRISTA stratospheric measurements, J. Geophys. Res., 106(D11), 12 253–12 265, 2001.
Ehhalt, D. H. and Tönnißen, A.: Hydrogen and carbon compounds
in the stratosphere, Proceedings of the NATO Advanced Study
Institute on Atmospheric Ozone: Its Variation and Human Influences, edited by: Aikin, A. C., Report No. FAA-EE-80-20, U.S.
Federal Aviation Administration, 1979.
Fish, D. J., Jones, R. L., and Strong, E. K.: Midlatitude observations
of the diurnal variation of stratospheric BrO, J. Geophys. Res.,
100(D9), 18 863–18 871, 1995.
Hendrick, F., Mueller, R., Sinnhuber, B.-M., Bruns, M., Burrows, J.
P., Chipperfield, M. P., Fonteyn, D., Richter, A., Van Roozendael,
M., and Wittrock, F.: Simulation of BrO Diurnal Variation and
BrO Slant Columns: Intercomparison Exercise Between Three
Model Packages, Proceedings of the 5th European Workshop
on Stratospheric Ozone, Saint Jean de Luz, France, 27 Sept.–
1 Oct. 1999, Air Pollution Research Report no. 73, European
Commission – DG XII, Brussels, 2000.
Hendrick, F., Barret, B., Van Roozendael, M., Boesch, H., Butz,
A., De Mazière, M., Goutail, F., Hermans, C., Lambert, J.-C.,
Pfeilsticker, K., and Pommereau, J.-P.: Retrieval of nitrogen
dioxide stratospheric profiles from ground-based zenith-sky UVvisible observations: validation of the technique through correlative comparisons, Atmos. Chem. Phys., 4, 2091–2106, 2004,
SRef-ID: 1680-7324/acp/2004-4-2091.
Heckel, A., Richter, A., Tarsu, T., Wittrock, F., Hak, C., Pundt, I.,
Junkermann, W., and Burrows, J. P.: MAX-DOAS measurements
of formaldehyde in the Po-Valley, Atmos. Chem. Phys., 5, 909–
918, 2005,
SRef-ID: 1680-7324/acp/2005-5-909.
Hönninger, G., Friedeburg, C. V., and Platt, U.: Multi Axis Differential Optical Absorption Spectroscopy (MAX-DOAS), Atmos.
Chem. Phys., 4, 231–254, 2004,
SRef-ID: 1680-7324/acp/2004-4-231.
Atmos. Chem. Phys., 6, 93–108, 2006
108
F. Hendrick et al.: Intercomparison exercise between radiative transfer models
Lenoble, J.: Radiative Transfer in Scattering and Absorbing Atmospheres: Standard computational procedures, A. Deepak Publishing, Hampton, Virginia USA, 1985.
Mayer, B. and Kylling, A.: Technical note: The libRadtran software
package for radiative transfer calculations – description and examples of use, Atmos. Chem. Phys., 5, 1855–1877, 2005,
SRef-ID: 1680-7324/acp/2005-5-1855.
Noxon, J. F., Whipple, E. C., and Hyde, R. S.: Stratospheric NO2 . 1.
Observational method and behaviour at midlatitudes, J. Geophys.
Res., 84, 5047–5076, 1979.
Petritoli, A., Ravegnani, F., Giovanelli, G., Bortoli, D., Bonafè, U.,
Kostadinov, I., and Oulanovsky, A.: Off-axis measurements of
atmospheric trace gases by use of an airborne ultraviolet-visible
spectrometer, Appl. Opt., 41(27), 5593–5599, 2002a.
Petritoli, A., Giovanelli, G., Kostadinov, I., Ravegnani, F., Bortoli,
D., Bonasoni, P., Evangelisti, F., Bonafè, U., and Calzolari, F.:
Tropospheric and stratospheric NO2 amount deduced by slant
column measurements at Mt. Cimone station, Adv. Space Res.,
29(11), 1691–1695, 2002b.
Platt, U: Differential optical absorption spectroscopy (DOAS), in:
Air Monitoring by Spectroscopic Techniques, Chem. Anal. Ser.,
edited by: Sigrist, M. W., John Wiley, New York, 127, 27–84,
1994.
Rozanov, A., Rozanov, V.-V., and Burrows, J.-P.: Combined
differential-integral approach for the radiation field computation
in a spherical shell atmosphere: Nonlimb geometry, J. Geophys.
Res., 105, D18, 22 937, 2000.
Rozanov, A., Rozanov, V.-V., and Burrows J.-P.: A numerical radiative transfer model for a spherical planetary atmosphere: combined differential-integral approach involving the Picard iterative
approximation, J. Quant. Spectrosc. Radiat. Transfer, 69, 491,
2001.
Rozanov, A., Rozanov, V., Buchwitz, M., Kokhanovsky, A., and
Burrows, J. P.: SCIATRAN 2.0. A new radiative transfer model
for geophysical applications in the 175–2400 nm spectral region,
Adv. Space Res., 36(5), 1015–1019, 2005.
Sarkissian, A., Roscoe, H. K., Fish, D., Van Roozendael, M., Gil,
M., Chen, H. B., Wang, P., Pommereau, J.-P., and Lenoble, J.:
Ozone and NO2 air-mass factors for zenith-sky spectrometers:
Intercomparison of calculations with different radiative transfer
models, Geophys. Res. Lett., 22(9), 1113–1116, 1995.
Schofield, R.: The vertical distribution of atmospheric BrO from
ground-based measurements, PhD thesis, University of Auckland, Auckland, 2003.
Schofield, R., Connor, B. J., Kreher, K., Johnston, P. V., and
Rodgers, C. D.: The retrieval of profile and chemical information from ground-based UV-visible spectroscopic measurements,
J. Quant. Spectrosc. Radiat. Transfer, 86, 115–131, 2004.
Shettle, E. P.: Models of aerosols, clouds, and precipitation for atmospheric propagation studies, AGARD Conference Proceedings No. 454: Atmospheric propagation in the UV, visible, IR
and mm-region and related system aspects, 1989.
Atmos. Chem. Phys., 6, 93–108, 2006
Sinnhuber, B.-M., Arlander, D. W., Bovensmann, H., Burrows, J.
P., Chipperfield, M. P., Enell, C.-F., Frieß, U., Hendrick, F.,
Johnston, P. V., Jones, R. L., Kreher, K., Mohamed Tahrin, N.,
Müller, R., Pfeilsticker, K., Platt, U., Pommereau, J.-P., Pundt, I.,
Richter, A., South, A. M., Tørnkvist, K.K., Van Roozendael, M.,
Wagner, T., and Wittrock, F.: Comparison of measurements and
model calculations of stratospheric bromine monoxide, J. Geophys. Res., 107(D19), 4398, doi:10.1029/2001JD000940, 2002.
Solomon, S., Schmeltekopf, A. L., and Sanders, R. W.: On the Interpretation of Zenith sky Absorption Measurements, J. Geophys.
Res., 92(D7), 8311–8319, 1987.
Spurr, R. J. D.: Linearized Radiative Transfer Theory: A general
discrete ordinate approach to the calculation of radiances and analytic weighting functions, with application to atmospheric remote sensing, PhD thesis, Technische Universiteit Eindhoven,
2001.
Stamnes, K., Tsay, S.-C., Wiscombe, W., and Jayaweera, K.: Numerically stable algorithm for discrete-ordinate-method radiative
transfer in multiple scattering and emitting layered media, Appl.
Opt., 27, 2505–2509, 1988.
Tørnkvist, K. K., Arlander, D. W., and Sinnhuber, B.-M.: Groundbased UV measurements of BrO and OClO over Ny-Ålesund
during Winter 1996 and 1997 and Andøya during Winter
1998/99, J. Atmos. Chem., 43(2), 75–106, 2002.
von Friedeburg, C.: Derivation of Trace Gas Information combining Differential Optical Absorption Spectroscopy with Radiative
Transfer Modelling, PhD thesis, University of Heidelberg, 2003.
Wagner, T., Dix, B., Friedburg, C. v., Frieß, U., Sanghavi, S., Sinreich, R., and Platt, U.: MAX-DOAS O4 measurements: A
new technique to derive information on atmospheric aerosolsPrinciples and information content, J. Geophys. Res., 109,
D22205, doi:10.1029/2004JD004904, 2004.
Wahner, A., Ravishankara, A. R., Sander, S. P., and Friedl, R.R.:
Absorption cross-section of BrO between 312 and 385 nm at 298
and 223 K, Chem. Phys. Lett., 152, 507–512, 1988.
Wahner, A., Tyndall, G. Y., and Ravishankara, A. R.: Absorption
cross-section for OClO as a function of temperature in the wavelength range from 240–490 nm, J. Phys. Chem., 91, 2735–2738,
1987.
Weidner, F., Bösch, H., Bovensmann, H., Burrows, J. P., Butz, A.,
Camy-Peyret, C., Dorf, M., Gerilowski, K., Gurlit, W., Platt, U.,
von Friedeburg, C., Wagner, T., and Pfeilsticker, K.: Balloonborne Limb profiling of UV/vis skylight radiances, ozone and
nitrogen dioxide: Technical set-up and validation of the method,
Atmos. Chem. Phys., 5, 1409–1422, 2005,
SRef-ID: 1680-7324/acp/2005-5-1409.
Wittrock, F., Oetjen, H., Richter, A., Fietkau, S., Medeke, T.,
Rozanov, A., and Burrows, J. P.: MAX-DOAS measurements
of atmospheric trace gases in Ny-Ålesund – Radiative transfer
studies and their application, Atmos. Chem. Phys., 4, 955–966,
2004,
SRef-ID: 1680-7324/acp/2004-4-955.
www.atmos-chem-phys.org/acp/6/93/